"If we misunderstand the client, we misspend 100 hours": Exploring conversational AI and response types for information elicitation

Abstract

Client-designer alignment is crucial to the success of design projects, yet little research has explored how digital technologies might influence this alignment. To address this gap, this paper presents a three-phase study investigating how digital systems can support requirements elicitation in professional design practice. Specifically, it examines how integrating a conversational agent and choice-based response formats into a digital elicitation tool affects early-stage client-designer collaboration. The first phase of the study inquired into the current practices of 10 design companies through semi-structured interviews, informing the system's design. The second phase evaluated the system using a 2x2 factorial design with 50 mock clients, quantifying the effects of conversational AI and response type on user experience and perceived preparedness. In phase three, the system was presented to seven of the original 10 companies to gather reflections on its value, limitations, and potential integration into practice. Findings show that both conversational AI and choice-based responses lead to lower dependability scores on the User Experience Questionnaire, yet result in client input with greater clarity. We contribute design implications for integrating conversational AI and choice-based responses into elicitation tools to support mutual understanding in early-stage client-designer collaboration.